Measurement and Modelling of Temperature Cold Pools in ......1 Measurement and Modelling of...
Transcript of Measurement and Modelling of Temperature Cold Pools in ......1 Measurement and Modelling of...
-
Measurement and Modelling of Temperature Cold Pools 1
in Cerdanya 2
M. Pagès1, N. Pepin
2 and J.R.Miró
1 3
4
1) Meteorological Service of Catalonia. Applied Research Department. C/Berlín 38-48, 4rt 08029 5
Barcelona. Spain 6
2) University of Portsmouth. Geography. Buckingham Building, Lion Terrace, Portsmouth, Hants, 7
PO1 3HE. United Kingdom 8
9
Correspondence to: [email protected] 10
11
Abstract 12
13
The Cerdanya valley in NE Spain experiences intense cold-air ponding (CAP) which decouples 14
the valley atmosphere from the regional circulation, especially in winter. This makes air 15
temperature prediction a challenge. A network of 40 temperature sensors was installed in 2012 16
along seven elevational transects to collect hourly temperatures throughout the cold pool, 17
enabling measurement of the detailed cold-pool structure for the first time. Sensors were also 18
installed in upper Conflent valley to the north-east for comparison, where previous research has 19
shown that there is reduced CAP. Sensor data is validated against AWS observations at two 20
locations. Through calculation of hourly lapse rates in various elevation bands for two years we 21
show frequent inversions developing up to 1450 m, and sometimes extending much higher than 22
this, concentrating in winter. Case studies of two intense episodes in December 2012 and 23
January 2013 show that model simulations, despite being able to simulate broad mechanisms of 24
CAP formation, underestimate the amount of cooling. This is due to over-enthusiastic simulation 25
of a low-level jet in the first case, and mountain waves in the second. Solving these model 26
problems has important consequences for future ability to predict episodes of extreme low 27
temperatures and associated hazards (frost, fog) in Cerdanya and mountain valleys elsewhere. 28
29
Key words: temperature inversion, lapse rates, complex terrain, mountain valley/basin, radiative 30
cooling, cold air pooling 31
32
33
34
35
36
Page 2 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
1 Introduction 37
The air temperature over complex terrain is governed by non-linear processes and is influenced 38
by both large-scale circulation of air-masses and local airflows; the interaction of airflow with 39
topography; the interplay between radiation and topographic shading; and variable surface 40
characteristics such as soil types, vegetation and presence/absence of snow, which influence 41
both energy and moisture exchange in the atmospheric boundary layer (ABL) (Barry, 2008). It is 42
challenging to characterize both thermal forcing (e.g. radiation balance) and subsequent mixing of 43
the air (turbulence) and thus predict temperature behaviour (Mahrt, 2006). 44
45
During nights dominated by high pressure, low winds and clear skies (especially in winter) the 46
surface temperature drops and the absence of extensive turbulence helps to grow a layer near 47
the surface with colder air than air above, producing a thermal inversion, i.e., the lapse rate of 48
temperature reverses in the lowest layer. This surface cold air flows downhill but can become 49
trapped by topography, a process known as cold air pooling (CAP) (Lundquist et al. 2008). Thus 50
numerous climatological studies of lapse rates within mountain areas have demonstrated clear 51
temporal and spatial variability (Rolland, 2003; Pepin et al. 1999). The advent of relatively 52
affordable stand-alone temperature sensors (Whiteman et al. 2000) now means that a denser 53
network of temperature measurements can be obtained than for most earlier studies, allowing 54
scientists to study the detailed spatial and temporal variation in vertical lapse rates in complex 55
terrain (e.g. Lewkowicz, 2008; Tang and Fang, 2006). Models have also been used to understand 56
the dynamics behind CAP formation and dissipation (Zängl et al. 2005; Collete et al. 2003). 57
Despite limitations in resolving boundary layer mechanisms (Lareau et al, 2013), meso-scale 58
model simulations have been shown to help understand physical processes that contribute to 59
persistent cold air episodes in confined valleys (Lu et al, 2014). 60
61
The aim of this research project is to understand the behaviour of air temperature in the basin of 62
Cerdanya in the eastern Pyrenees, (~15 km wide and ~30 km long) using such a dense network 63
of temperature sensors, and to assess the extent to which ARPS model simulations can 64
represent this temperature variability, especially during extreme and persistent inversion events. 65
Section 2 sets the study in the context of past research. Section 3 outlines the methods, both 66
field-based and modelling. Section 4 presents preliminary field results, providing insight on the 67
climatology of inversions within Cerdanya. Section 5 evaluates the success of the ARPS model in 68
reproducing observed temperature patterns using two contrasting strong inversion episodes as 69
test cases. This is followed by a discussion of our findings. 70
71
2 Physical Theory and Background Rationale 72
Under stable conditions the lack of turbulence means that cold air becomes trapped in mountain 73
valleys and basins and does not mix with air above until displaced/dispersed via mechanical 74
Page 3 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
and/or thermal mechanisms (Whiteman, 1982). The air layer is said to become “decoupled” from 75
the free atmosphere. Forecasting temperature fields, especially in valley bottoms, is complicated 76
by a range of physical processes. In contrast, temperatures at exposed summit locations are 77
strongly correlated with free-atmospheric temperatures at the same elevation (McCutchan, 1983; 78
Pagès and Miró, 2010; Pepin and Seidel, 2005). 79
80
Field studies examining CAP have been carried out in numerous mountain ranges including the 81
Alps (Steinacker et al, 2007; Whiteman et al, 2004), and the western U.S. (Clements et al. 2003; 82
Whiteman et al., 1999). Many studies have been concerned with the effect of CAP (and 83
atmospheric stability in general) on air quality and possible trapping of pollutants, particularly in 84
the Alps (Chazette et al. 2005: Chemel et al. 2016). Long-term campaigns in mountainous terrain 85
are often limited by challenging meteorological conditions so most past studies have 86
concentrated on short-term diurnal forcing including morning dissipation (Whiteman 1982, 87
Whiteman et al. 1999) and evening transitions in summer convective situations (Nadeau et al. 88
2013). Only a few studies have investigated CAP evolution over several days (Lareau et al. 89
2013), partly because persistence of CAP beyond one day is relatively uncommon, at least in 90
mid-latitudes. From December 2010 until February 2011 an intensive campaign was carried out in 91
Salt Lake City Valley (Utah), to detect persistent CAP, providing an important database covering 92
longer events (Lareau et al, 2013; Lu et al. 2014). 93
94
Understanding both the mechanisms by which decoupling of the surface inversion layer is 95
initiated and how it is destroyed will help improve knowledge of the physical mechanisms of the 96
ABL and improve parameterizations of such processes in weather forecasting models (Whiteman, 97
1982). Detailed air temperature measurements help scientists to understand and calibrate 98
particular physical parameterizations used by models or to test the ability of different model 99
configurations (vertical levels, grid size, etc) in reproducing realistic temperature patterns. In 100
general, the resolution needed for numerical models to simulate the behaviour of temperature in 101
complex terrain is around tens of meters (Chow et al, 2006). This takes into account fine-scale 102
turbulence that transfers cold air from the surface to the atmosphere. However such simulations, 103
often employed in research models at such a high resolution, are very expensive computationally 104
and in Numerical Weather Prediction (NWP) the finest scale phenomena are often approximated 105
using parameterizations based on downscaling from the coarser model scale. Such downscaling 106
relies on relationships calculated in laboratories or experimental campaigns, and relationships 107
have to be adjusted to each location where they are applied (Cuxart et al. 2000). 108
109
The mesoscale context which links synoptic forcing with microscale mechanisms is critical to 110
understanding both CAP formation and breakup (Lareau et al 2015a). Traditional radiative 111
analysis e.g. the role of Topographic Amplification Factor (TAF) in controlling radiation loss and 112
Page 4 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
downstream flow of air (Steinacker, 1984; McKee and O’Neal 1989) is often only part of the 113
explanation. Low Level Jets (LLJ) (Cuxart and Jimenez, 2007) or mountain gravity waves (Lee et 114
al, 1989) can generate turbulence through shear and this produces “turbulent erosion” which 115
weakens any temperature inversion (Lareau et al 2015b). Mesoscale numerical models are thus 116
a useful tool to study such complex terrain meteorology (Fast and Darby, 2004; Chow et al, 2006) 117
and the relationship between turbulent transfer and persistence of CAP (Rakovec et al, 2002; 118
Zangl, 2005). 119
120
From a climate perspective, relating detailed temperature readings in areas of complex 121
topography to synoptic conditions also allows an understanding of how temperatures may change 122
in a warmer climate (Lundquist and Cayan, 2007) and an assessment of how climate change 123
could influence patterns of decoupling and consequently magnify or reduce temperature trends in 124
particular topographic locations (Daly et al. 2010). Previous work examining decoupling in a 125
range of latitudinal zones from the Arctic to the tropics (Gustavsson et al. 1998; Chung et al. 126
2006; Daly et al. 2003) has shown that different synoptic controls are important in different 127
locations. For example, the interaction between CAP and synoptic forcing (as measured by an 128
anticyclonicity index) is very strong in the northern states of the United States (Pepin et al. 2011) 129
and Oregon (Daly et al. 2010), but weak in the tropics and southern U.S. 130
131
CAP is often critical to ecological systems, defining refugia for example (Dobrowski, 2011). Work 132
in Finland has shown that topographically controlled extreme low minimum temperatures (
-
Figure 1) is 15 km wide and flat bottomed, with the valley bottom averaging around 1000 m asl. 151
To the north and the south mountain ranges rise to over 2000 m asl. To the west (downstream) 152
the Segre flows into a narrow gorge which provides a constriction for down-valley flow. To the 153
east the land rises to the Col de Perche (1500 m asl) before dropping away into Conflent which 154
flows north-east to the Mediterreanean. The upper (eastern) part of Cerdanya is particularly prone 155
to temperature inversions (Pepin & Kidd, 2006). Due to surrounding shelter, the inner valley areas 156
are semi-arid with precipitation of only 600 mm/year, but in the surrounding mountains the annual 157
total reaches over 1400 mm/year (Xercavins, 1985). Convective precipitation is common in 158
summer. The Cerdanya region has two automatic weather stations (AWS) which we use for 159
calibration, operated by the Meteorological service of Catalonia (Das, 1500m, Cadi Nord, 2200 m) 160
(Figure 1). 161
162
b. Field data collection 163
A network of 40 HOBO U23-001 sensors (temperature/relative humidity) was installed across the 164
valley (see Figure 1). Data was recorded every 30 minutes from July 2012 until July 2014. The 165
accuracy of the temperature sensor is +/-0.21ºC and the operating range from -40ºC to 70ºC, 166
adequate for local scale temperature monitoring (Whiteman et al. 2000). The network includes 167
contrasting mountain slopes with differential aspects and land-use. Differences in incident 168
radiation could affect the valley temperature structure and cause asymmetry in any thermally 169
driven circulation system (Whiteman et al. 1989). 170
171
Seven main transects were installed, three each on the north and south side of Cerdanya, and 172
one in Conflent. Sensors were installed at approximately equal elevational intervals (every 200-173
250 metres), to include a difference of around 1000 m on all transects. An additional transect 174
along the valley bottom encompassed both the gorge in the lower reaches (Val1-Val3) and the 175
more open upper valley basin (Val4 and Val5). Based on previous work (Pepin and Kidd, 2006) 176
which identified that the temperature regime in Conflent was decoupled from that in Cerdanya, 177
the network included an extra transect in Conflent. Conflent’s behaviour is more similar to the free 178
atmosphere with limited cold air drainage. All transects covered a similar elevation range, with the 179
highest sites around or slightly above treeline (~2100-2400 m). Sites were chosen so that local 180
aspect was representative of the broad slope (e.g. south facing on the Malniu transect) and 181
topographic hollows with microclimates were avoided, except in the valley bottom. Freely draining 182
slope sites with relatively open forest cover were preferred. 183
184
To protect sensors from solar and longwave radiation, the technique followed by Pepin et al. 185
(2010) was used. Sensors were placed into white PVC pipes of 30-40 cm length and a diameter 186
of 15 cm, to protect from direct radiation, yet still allow sufficient ventilation. PVC pipes were 187
attached to trees (36 cases) or other permanent fixtures (4 cases) at a height of 1.5 m above 188
Page 6 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
ground level with the top end facing north at an angle of 45 degrees. The majority of evergreen 189
trees chosen were Pine (Pinus alepensis and Pinus sylvestris) but there were also some Juniper 190
(Juniperus phoenicea). The screening method has been tested in a variety of environments, 191
including one of the most extreme radiative environments on Earth on the ice fields of Kilimanjaro 192
(Duane et al. 2008) and has been shown to be effective at preventing overheating. Although trees 193
can create their own microclimate, large parts of the study area are forested so use of trees 194
allows consistency. Mean differences between sensors in trees and those in a standard screen 195
were shown to be negligible (see section 4a). 196
197
The elevational distribution of sensors ranges from 739 m (Val1) to 2484 m (Mas1). Because of 198
some faulty dataloggers, some limited data is missing. Screen height was designed to avoid snow 199
outside of drift areas. However, build-up of snow around the sensor occurred at two sites in 200
February/March (Les1 and Cad1) and such data was deleted (about 4 months of data in total). 201
Nevertheless we have a fairly comprehensive dataset of 2 years of data at 39 locations. One 202
sensor (Eyn5) was stolen. 203
204
c. Mesoscale model 205
To understand the mechanisms of CAP formation and evolution the mesoscale model ARPS 206
(Advanced Regional Prediction System), developed at the Center for Analysis and Prediction of 207
Storms at the University of Oklahoma, is employed. ARPS is intended for mesoscale and small-208
scale atmospheric simulations (Chow et al. 2006). A detailed model description is found in Xue et 209
al. (2000). ARPS is a non hydrostatic, compressible large-eddy simulation code written for 210
mesoscale and fine-scale atmospheric flows. A Fourth-order scheme is used for advection terms 211
and temporal discretization is performed in order to avoid numerical instabilities due to high-212
frequency acoustic waves. For boundary layer parametrization a 1.5th order turbulent kinetic 213
energy (TKE) turbulence closure is used. Turbulent sensible heat, latent heat, and momentum 214
fluxes in the surface layer are based on the similarity theory of Monin and Obukhov (1954). The 215
model was run in a one-way nested mode. The outermost simulation had 27 km grid spacing 216
intialized with the GFS model, and this was successively nested down with a factor of 1:3 to a 1 217
km grid. The latter is used to study the mechanisms of CAP formation in two case studies. 218
219
A 1 km grid was chosen in order to make model topography as close as possible to reality, 220
improving the representation of mountain related phenomena (Warner et al, 1997). The vertical 221
levels of the ARPS model use a coordinate system that follows the terrain (Xue et al, 2000). A 222
fine vertical interval near the ground is required to solve turbulent and radiative exchanges, but in 223
zones with complex terrain the ratio between the horizontal and vertical grid spacings (∆x/∆z) 224
must not be too large as this creates numerical instabilities due to error propagation (Chow et al. 225
Page 7 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
2006). Therefore a variable vertical resolution was employed, stretching from the surface 226
upwards using a hyperbolic tangent function from 30 m grid-spacing near the ground up to 400 m 227
grid-spacing at 14000 m asl. 228
229
4. Results 230
a. Calibration of sensor temperatures against AWS 231
Observed temperatures from the sensor network were compared with those from longer-term 232
AWS at a) Cad1 (Cadi Nord AWS) near the treeline on the north-facing aspect of the Cadi range 233
(2229 m) and, b) Val4 (Das AWS) in the valley bottom (1109 m). 234
235
The difference between temperatures at the Cadi Nord AWS sensor and the Cad1 data logger 236
sometimes reached >5˚C in late winter/spring caused by snow build up around the datalogger 237
which insulated it. Ignoring such brief snowy periods at Cad1, the mean difference is only -0.39˚C 238
(logger colder). This is expected since the logger is about 30-40 m higher in elevation than the 239
AWS at Cad1. Differences between the Val4 sensor and Das AWS were also small and more 240
consistent. A seasonal and diurnal breakdown of temperature differences at Das showed the 241
mean difference as 0.75˚C (logger warmer) but this difference was smaller during the day and in 242
spring/summer. Any difference could relate at least in part to differences in screening (the shelter) 243
as well as equipment (sensor differences). However the reduced difference during the day (when 244
radiative errors are potentially larger) suggests that screening differences are relatively 245
unimportant. Overall the differences are small, and the correlation between instantaneous values 246
of temperature is extremely high (r=0.998, n=29,026). The analysis therefore gives high 247
confidence in datalogger performance. 248
249
b. Lapse rates 250
Lapse rates were calculated using the gradient of the ordinary least squares regression line fitted 251
to temperatures against elevation for all 39 stations and for stations in three elevational bands 252
(below 1450 m; 1450-1900 m; and above 1900 m) (13 stations in each to facilitate comparison). 253
They were also calculated separately for the Conflent transect (6 stations from Fon0 to Fon5). 254
Positive lapse rates represent inversion conditions. Table 1 lists mean lapse rates and other 255
statistics based on calculations every 6 hours (n~2900). In Conflent there is a subdued seasonal 256
lapse rate cycle, which peaks in spring (-7.02ºCkm-1
in March) and is weakest in August (-5.51 257
ºCkm-1
). Inversions are rare, as indicated by the low frequencies in the positive lapse rate column 258
(>0ºCkm-1
). Superadiabatic lapse rates (-4 261
ºCkm-1
), influenced by frequent absolute inversion episodes (>20 % of time in December). The 262
Page 8 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
complexity of the temperature structure for the dataset as a whole is also illustrated in winter by 263
lower mean r2 values for elevation vs temperature (
-
scatter in temperatures as a result of slope aspect differences, particularly in the low elevation 301
band. A discussion of such microclimate influences is beyond the scope of this paper. 302
303
Using significant codings only (right hand column of Table 2), isolated low level inversions in the 304
lowest layer only (category 1) are the most common (86% of all inversions). An analysis of the 305
frequency distribution of shallow layer inversions (category 1) according to time of day and 306
season (table not shown) shows a peak in winter (86%), and at 0600 UTC (38%). However there 307
is a significant proportion (38%) remaining at 1200 and 1800 UTC in January/December. All the 308
other more vertically extensive inversions extending up to and including mid and or high levels 309
(codes 2 to 5) are confined solely to winter at 0000 or 0600 UTC. Thus inversions covering more 310
than the bottom layer are considerably less common than shallow ones, and their timing/build-up 311
requires long periods of negative energy balance. Two unusual cases were recorded in July when 312
an inversion formed in the upper band (above 1900 m) but not in the lower valley (coded as 4). 313
Although these were still recorded at night, the summer maximum suggests synoptic forcing may 314
be partly responsible, upper level subsidence inversions being typical of summer anticyclones. 315
316
The most persistent inversions were recorded in the lowest layer (
-
difference between transects. The Fontpedrouse transect (outside Cerdanya) shows relative 339
instability in the lowest layers in comparison with the other transects, particularly in spring and 340
summer (top row). This demonstrates a decoupling of the Cerdanya atmosphere at the lower 341
levels from the surrounding region (represented by Fontpedrouse). The profiles for the four 342
transects within Cerdanya are broadly similar, showing strong stability in the lowest layers in 343
autumn and winter. In spring and summer Masella develops some limited instability below 1500 344
m asl but this may result from microclimate differences between the two lowest transect points. 345
346
5. Modelling: case studies of intense inversions 347
To understand the mechanisms of CAP formation, two case studies were chosen: 11-12 348
December 2012 and 28-30 January 2013. These were two of the most intense inversions 349
recorded (as measured by lapse rate gradients). They also followed different progressions and 350
had contrasting synoptic forcing so provide a good insight into model strengths and deficiencies. 351
The first case develops following cold air advection under clear skies and weak winds, and the 352
second case corresponds with upper level warm advection that strengthens an existing low level 353
inversion (Mahrt, 1999). Model predictions of the temperature field were compared with observed 354
temperatures at all 39 logger sites, to assess how well the forecast models reproduce the 355
temperature patterns. Table 3 shows bias statistics including RMSE for both simulations at a 356
variety of sites. In general model bias is positive at the lowest stations (model too warm) 357
suggesting an underestimation of CAP. 358
359
Case 1: 11-12 December 2012 360
The inversion (Figure 5a) developed first at a low level (below 1450m) in the early hours of 11 361
December and strengthened to over 5ºC/km by 0600 UTC. By 0600 UTC on 12 December the 362
inversion had risen to include the middle elevation band (up to 1900 m), and a positive lapse rate 363
was recorded for the 39 stations as a whole. An anticyclone was centred over France by 0000 364
UTC on 12 December (meaning weak winds and clear skies) although cold air advection had 365
been prevalent the previous day (Figure 6 top row). By 1800 UTC on 12 December the inversion 366
had all but disappeared. Nevertheless the valley still shows a relatively stable profile up to the 367
middle levels. 368
369
Although the ARPS model does simulate CAP there are some large differences between 370
modelled and observed temperature patterns (Figure 7). While the model is quite successful at 371
Les4 (on the south facing aspect), it is much poorer at valley bottom sites (Val3) which are on 372
occasions more than 5ºC colder than predicted by the model. Warm air advection at the highest 373
elevation site (Fon0) which would strengthen the inversion is also not modelled well, although the 374
general temporal increase is similar for modelled and observed temperatures. 375
376
Page 11 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
A map of the ARPS simulations (Figure 8a) shows that at 0000 UTC on 12 December cold air 377
flows downhill from the high area around the Col de la Perche (which cools quickly after sunset) 378
to the Cerdanya valley (left ellipse) and to Conflent valley (right ellipse). In the Cerdanya valley 379
the air drains downhill (westward) until the Segre flows into a constriction where the wind speed 380
slows and the air remains quiet (Figure 8b). This helps efficient cooling and formation of the cold 381
pool. On the other hand, drainage to the east within Conflent is strong because it is both steep 382
and narrow. The map also shows decoupling between north-west winds at the mountain top level 383
and the valley bottom. 384
385
The cross-section along the valley bottom (Figure 8b) shows how a temperature inversion 386
appears near the ground. About 200 m above the ground is a low level jet, possibly a result of the 387
temperature gradient produced by heating differences in sloping terrain (Holton, 1967). This low 388
level jet (LLJ) has been known to produce model turbulence that mixes the air and erodes the 389
inversion (Cuxart and Jimenez, 2007). So, one cause of model temperature positive bias at the 390
bottom valley sensors (model too warm) could be related to the overestimation of this LLJ or its 391
impacts. 392
393
Figure 8c shows a weak temperature inversion on the Masella transect at 0000 UTC. This 394
remains broadly similar at 0600 UTC (Figure 8d) with fairly uniform temperatures up to ~2000 m 395
despite the potential temperature profile at 0600 UTC showing a very stable boundary layer 396
(Figure 8e). 397
398
Case 2: 29-31 January 2013 399
In this example (Figure 5b) inversions were more intense and longer-lasting. The inversion in the 400
lowest band (1900 m). The model was partly able to simulate CAP 405
formation but shows some large differences with the observations (Figure 9). Both strong upper-406
level warm air advection at Fon0 which raised temperatures by over 15C between 1200 UTC on 407
28 and 30 January (which strengthened the inversion) and valley bottom CAP at Val3 are again 408
not simulated effectively. During the day the valley sites warm up more than in the model and 409
overall the model has a cold bias at most valley bottom locations (despite the nocturnal warm 410
bias at Val3). 411
412
Although upper level synoptic flow is from the north-west (Figure 10a), these winds are channeled 413
to flow through the Col de la Perche from the south-west. Thus this channeling acts to minimise 414
Page 12 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
any local drainage flow from the north-east into Cerdanya. Figure 10b shows that after sunset 415
drainage west from the Col de la Perche at ground level is very weak (unlike in the previous case) 416
and no low level jet appears above the valley bottom because there is lack of a significant vertical 417
temperature gradient. High Cerdanya is much more exposed to wind than the Masella transect so 418
only the latter is able to retain a weak inversion near the surface at 0000 UTC (Figure 10c). By 419
0600 UTC this inversion is shallow and strong winds are reported across the valley down to 420
around 1600 m (Figure 10d). The model winds over the mountains produce mountain gravity 421
waves and turbulence as shown in the cross-section of wind and potential temperature across 422
Masella (Figure 10e) which erodes the earlier inversion that appeared at 0000 UTC. Mountain 423
gravity waves can become intense when the air is stable and well stratified (Vosper, 2004) and 424
are possibly responsible for the erosion of CAP in the model in this case. Observed temperatures 425
however show that in this case study there was no CAP destruction. Sometimes CAP acts as an 426
inhibitor of mountain waves penetrating more incised valleys which prevents mixing (Zhong et al, 427
2001). Thus the cause of model inaccuracy in this case is probably overestimation of mountain 428
waves and their associated turbulence. 429
430
6. Discussion 431
The wide variation in lapse rates fits with other studies, with steeper rates common in summer 432
and during the day, and weaker rates at night and in winter due to negative radiation balance 433
(Rolland, 2003; Minder et al. 2010). The spring peak in lapse rate has been found in numerous 434
other mid-latitude studies (Minder et al. 2010; Pepin, 2000), thought to be partly a response to a 435
strong elevational gradient in snow cover, but also an air-mass effect where the surface starts to 436
heat but the upper air still remains cold. 437
438
The daily cycle in lapse rates with stronger rates during the day, and weaker mean rates at night 439
is similar to Blandford et al. (2008); Bolstad et al. (1998); and numerous other studies. Inversions 440
are therefore most common at night and are strongest in the lowest elevation band (below 1450 441
m) where the topography forms a constriction. Unexpectedly the steepest daytime lapse rates 442
and the strongest diurnal signal in lapse rates were recorded in the highest elevation band (above 443
1900 m). This coincides with the forest/alpine tundra ecotone and the change in vegetation may 444
well enhance vertical lapse rates near the treeline under conditions of intense sunlight and snow 445
cover (see Pike et al. 2013). However other studies in the tropics have shown that the treeline 446
can weaken the near surface lapse rate on an annual basis (Pepin et al. 2010) and more detailed 447
work is required to examine the effect of land-use changes on lapse rate structure. 448
449
Using potential temperatures immediately showed up differences in stability. Conflent shows a 450
differential profile, as may be expected in a freely draining V-shaped valley with steep longitudinal 451
gradient and limited propensity for CAP (McKee and O’Neill 1989; Lundquist et al. 2008). The sky 452
Page 13 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
view factor is also reduced particularly in the lower reaches of Conflent, which might reduce 453
radiative losses and prevent temperatures from falling as low in the valley bottom. Whiteman et 454
al. (2014) suggest this as a possible mechanism in some depressions and sinkholes which 455
remain warmer and have slightly less stable temperature structures than their surroundings. 456
Cerdanya in contrast shows enhanced stability in the lower layers. 457
458
We examined relationships between lapse rates and synoptic indices (as represented by flow 459
strength values and an A-C index: see Daly et al. 2010) but the results were not strong (therefore 460
not shown). Preliminary analysis showed that sometimes a decoupling of the free atmosphere 461
and the valley circulation produced cold air ponding in the valley bottom even when upper level 462
winds remained strong. A good example of this is in the second case study when the upper level 463
jet stream was sited not far north of Cerdanya (Figure 8: bottom left). The reasons for this lack of 464
direct synoptic control require more research. 465
466
Model simulations of two case studies show variable success at simulating the cold pool. In both 467
cases model bias was too warm in the valley bottom and mean bias was negatively correlated 468
with elevation. In contrast, bias was unrelated to elevation error (model minus station elevation - 469
not shown) even though elevation errors could be substantial at individual sites (up to nearly 500 470
m in the worst case). Thus a correction for model elevation error alone would not be enough to 471
explain model errors. In the first case study there was strong cold advection following an 472
anticyclone (which was then trapped by the topography), whereas in the second case there was 473
warm advection over the top of a colder valley bottom. In both cases the model underestimated 474
the strength and persistence of the cold pool. In the first case the overestimation of a low-level jet 475
and in the second case the formation of unstable mountain waves, led to increased turbulence 476
near the ground which mitigated the cold air pool. 477
478
7. Conclusions and future work 479
Temperature mapping of Cerdanya in the eastern Pyrenees illustrates the existence of inversions 480
which often form in winter and at night. This is in contrast to Conflent to the east which shows a 481
more similar regime to the free-air. More detailed analysis is required of the lower elevations 482
(below 1450 m) to illustrate exactly where the cold air drainage is most intense and/or prolonged. 483
Synoptic forcing has been shown to be relatively weak, and intense inversions can occur even 484
when upper winds are fairly strong. Perhaps because of this, the ARPS model is not always able 485
to simulate the extent and severity of the cold air drainage, and as such over-predicts 486
temperatures at prone locations under such circumstances as mountain wave development. 487
488
Research is attempting to improve temperature predictions in the complex terrain of Cerdanya, in 489
particular through improving the parameterisation of land-use in the WRF model (Esteve pers. 490
Page 14 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
Comm. 2015). Effective simulation of temperature inversions is particularly important to 491
incorporate in future analyses. The Cerdanya valley does not have a uniform cross-section and 492
there are wide and narrow zones. Downstream variation in the Topographic Amplification Factor 493
(TAF) (Steinacker, 1984) is therefore expected to cause alternate zones of cold air pooling and 494
draining. Although we have illustrated the presence of inversions for Cerdanya using transects 495
upstream of constrictions (in particular Masella and Malniu transects), without more detailed 496
sensor patterns across the valley bottom we cannot quantify the exact location of the lowest 497
temperatures. Of current sensors, Val3 consistently shows the lowest minima, just upstream of 498
the narrowing of the valley east of the town of Bellver. More sensors at higher spatial resolution 499
were installed in July 2014 to investigate whether this is a localised cold-pool or more extensive. 500
501
Finally there is need to apply information on inversion occurrence to examine how future changes 502
in atmospheric circulation, and in free atmospheric characteristics, may change the propensity 503
towards inversion formation in future climate scenarios. This will enable more realistic 504
downscaled predictions about climate warming in isolated basins such as Cerdanya. 505
506
Acknowledgements 507
Funding for equipment (temperature sensors) was provided by the University of Portsmouth, and 508 funding for fieldwork was partly provided by the Meteorological Service of Catalonia. We thank 509 the reviewers for providing constructive comments which improved the manuscript. 510 511 References 512 513 Barry, RG. 2008. Mountain Weather and Climate, Cambridge University Press, Cambridge, UK. 514
ISBN 978-0-521-86295-0. 515 516 Blandford T, Humes K, Harshburger B, Moore B, Walden V, Ye H. 2008. Seasonal and synoptic 517 variations in near surface air temperature lapse rates in a mountainous basin. J. Appl. Meteorol. 518
Climatol., 47(1), 249–261, doi:10.1175/2007JAMC1565.1. 519 520 Bolstad PV, Swift L, Collins F, Regniere J. 1998. Measured and predicted air temperatures at 521
basin to regional scales in the southern Appalachian mountains. Agric. For. Meteorol., 91(3––––4), 522
161–176. 523 524 Chazette P, Couvert P, Randriamiarisoa H, Sanak J, Bonsang B, Moral P, Berthier S, Salanave 525 S, Toussaint F. 2005. Three-dimensional survey of pollution during winter in French Alps valleys. 526 Atmos. Env. 39, 1035-1047. 527 528 Chemel C, Arduini G, Staquet C, Largeron Y, Legain D, Tzanos D, Paci A. 2016. Valley heat 529 deficit as a bulk measure of wintertime particulate air pollution in the Arve River Valley. Atmos. 530 Env. 128, 208-215. 531 532 Clements CB, Whiteman CD, Horel JD. 2003. Cold-air-pool structure and evolution in a mountain 533 basin: Peter Sinks, Utah. J. Appl. Meteorol. 42: 752–768. 534 535
Page 15 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
Chow FK, Weigel AP, Street RL, Rotach MW, Xue M. 2006. High-resolution large-eddy 536 simulations of flow in a steep Alpine valley. Part I: Methodology, verification, and sensitivity 537 studies. J. Appl. Meteor. Climatol., 45, 63–86. 538 539 Chung U, Seo HH, Hwang KH, Hwang BS, Choi JJ, Lee T, Yun JI. 2006. Minimum temperature 540 mapping over complex terrain by estimating cold air accumulation potential. Agric. For. Meteorol. 541 137: 15–24. 542 543 Colette A, Chow FK, Street RL. 2003. A numerical study of inversion-layer breakup and the 544 effects of topographic shading in idealized valleys. J Appl. Meteorol. 42(9), 1255-1272. 545 546 Cuxart J, Bougeault P, Redelsperger JL. 2000. A turbulence scheme allowing for mesoscale and 547 large-eddy simulations. Quarterly Journal of the Royal Meteorological Society, 126(562), 1-30. 548 549 Cuxart J, Jiménez MA. 2007. Mixing processes in a nocturnal low-level jet: An LES study. Journal 550 of the atmospheric sciences, 64(5), 1666-1679. 551 552 Daly C, Helmer EH, Quinones M. 2003. Mapping the climate of Puerto Rico, Vieques, and 553 Culebra. Int. J. Climatol. 23: 1359–1381. 554 555 Daly C, Conklin DR, Unsworth MH. 2010. Local atmospheric decoupling in complex topography 556
alters climate change impacts. Int.J. Climatol., 30, 1857–1864. 557 558 Dobrowski SZ. 2011. A climatic basis for microrefugia: The influence of terrain on climate. Global 559 Change Biology 17: 1022–1035. doi:10.1111/j.1365-2486.2010.02263.x. 560 561 Duane W, Pepin NC, Losleben ML, Hardy DR. 2008. General characteristics of temperature and 562 humidity variability on Kilimanjaro, Tanzania. Arct. Antarc. and Alp. Res. 40 (2), 323-334. 563 564 Esteve BJ. 2015. pers. comm. Land use influence in WRF model. A high resolution mesoscale 565 modelling over Oriental Pyrenees, unpub. MSc Thesis, Universitat de Barcelona, Diagonal 645, 566 Barcelona, Spain. 567 568 Fast JD, Darby LS. 2004. An Evaluation of Mesoscale Model Predictions of Down-Valley and 569 Canyon Flows and Their Consequences Using Doppler Lidar Measurements during VTMX 2000. 570 J. Appl. Meteor., 43, 420–436. 571 572 Gustavsson T, Karlsson M, Bogren J, Lindqvist S. 1998. Development of temperature patterns 573 during clear nights. J Appl. Meteorol. 37: 559–571. 574 575 Holton JR. 1967. The diurnal boundary layer wind oscillation above sloping terrain. Tellus, 19(2), 576 199-205. 577 578 Lareau NP, Crosman E, Whiteman CD, Horel JD, Hoch SW, Brown WO, Horst TW. 2013. The 579 persistent cold-air pool study. Bulletin of the American Meteorological Society, 94(1), 51-63. 580 581 Lareau NP, Horel JD. 2015a. Dynamically induced displacements of a persistent cold-air pool. 582 Boundary-Layer Meteorology, 154(2), 291-316. 583 584 Lareau NP, Horel JD. 2015b. Turbulent erosion of persistent cold air pools: numerical 585 simulations. Jnl Atmos. Sciences, 72, 1409-1427. 586 587 Lee TJ, Pielke RA, Kessler RC, Weaver J. 1989. Influence of cold pools downstream of mountain 588 barriers on downslope winds and flushing. Mon. Weather Rev., 117(9), 2041-2058. 589 590
Page 16 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
Lewkowicz AG. 2008. Evaluation of miniature temperature-loggers to monitor snowpack evolution 591 at mountain permafrost sites, Northwestern Canada. Permafrost and Periglacial Processes 19: 592 323-331. 593 594 Lu W, Zhong S. 2014. A numerical study of a persistent cold air pool episode in the Salt Lake 595 Valley, Utah. Journal of Geophysical Research: Atmospheres, 119(4), 1733-1752. 596 597 Lundquist JD, Cayan DR. 2007. Surface temperature patterns in complex terrain: Daily variations 598 and long‐term change in the central Sierra Nevada, California. J. Geophys. Res. Atmos. 112, 599 D11124. 600 601 Lundquist JD, Pepin N, Rochford C. 2008. Automated algorithm for mapping regions of cold-air 602 pooling in complex terrain. J. Geophys. Res., 113, D22107, doi:10.1029/ 2008JD009879. 603 604 Mahrt, L. 1999. Stratified Atmospheric Boundary Layers, Boundary-Layer Meteorol. 90, 375–396. 605 606 Mahrt L. 2006. Variation of Surface Air Temperature in Complex Terrain. J. Appl. Meteor. 607 Climatol. 45, 1481–1493. 608 609 McCutchan MH. 1983. Comparing Temperature and Humidity on a Mountain Slope and in the 610 Free Air Nearby. Mon. Wea. Rev. 111, 836–845. 611 612 McKee TB, O’Neal RD. 1989. The role of valley geometry and energy budget in the formation of 613 nocturnal valley winds. J. Appl. Meteorol., 28: 445–456, doi:10.1175/1520-0450(1989)0282.0.CO;2. 615 616 Minder JR, Mote PW, Lundquist JD. 2010. Surface temperature lapse rates over complex terrain: 617 lessons from the Cascade Mountains. J. Geophys Res 115:D14122. doi:10.1029/2009JD013493. 618 619 Monin AS, Obukhov, AM. 1954. Osnovnye zakonomernosti turbulentnogo peremeshivanija v 620 prizemnom sloe atmosfery (Basic Laws of Turbulent Mixing in the Atmosphere Near the Ground). 621 Trudy Geofiz. Inst. AN SSSR 24(151), 163–187. 622 623 Nadeau DF, Pardyjak ER, Higgins CW, Huwald H, Parlange MB. 2013. Flow during the evening 624 transition over steep Alpine slopes. Quart. Jnl. Roy. Met. Soc. 139, 607-624. 625 626 Pagès M, Miró, JR. 2010. Determining temperature lapse rates over mountain slopes using 627 vertically weighted regression: a case study from the Pyrenees. Meteorol. Appl., 17, 53-63, doi: 628 10.1002/met.160. 629 630 Pepin, NC. 2000. Twentieth century change in the climate record for the Front Range, Colorado, 631 U.S.A. Arct. Antarc. Alp. Res. 32(2): 135-146. 632 633 Pepin N, Benham D, Taylor K. 1999. Modeling lapse rates in the maritime uplands of northern 634 England: Implications for climate change. Arct. Antarct. Alp. Res. 31,151-164. 635 636 Pepin N, Daly C, Lundquist J. 2011, The influence of surface versus free-air decoupling on 637 temperature trend patterns in the western United States. J. Geophys. Res. Atmos., 116 (D10109). 638 16pp. ISSN 0148-0227 10.1029/2010JD014769. 639 640 Pepin NC, Duane W, Hardy D. 2010. The montane circulation on Kilimanjaro, Tanzania, and its 641 relevance for the summit ice-fields: Comparison of surface mountain climate with equivalent 642 reanalysis parameters. Glob. Planet. Change 74, 61-75. 643 644 Pepin N, Kidd D. 2006. Spatial temperature variation in the Eastern Pyrenees. Weather 61, 300-645 310. 646
Page 17 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
647 Pepin NC, Seidel DJ. 2005. A global comparison of surface and free-air temperatures at high 648 elevations, J. Geophys. Res., 110, D03104, doi:10.1029/ 2004JD005047. 649 650 Pike G, Pepin N, Schaefer M. 2013. High latitude local scale temperature complexity: the 651 example of Kevo Valley, Finnish Lapland. Int. J. of Climatol. doi: 10.1002/joc.3573. 652 653 Rakovec J, Merše J, Jernej S, Paradiž B. 2002. Turbulent dissipation of the CAP in a basin: 654 Comparison of observed and simulated development, Meteorol. Atmos. Phys., 79(3–4), 195-213. 655 656 Rolland C. 2003. Spatial and Seasonal Variations of Air Temperature Lapse Rates in Alpine 657 Regions. J. Climate, 16, 1032–1046. 658 659 Steinacker R. 1984. Area–height distribution of a valley and its relation to the valley wind. Beitr. 660 Phys. Atmos. 57, 64–71. 661 662 Steinacker R, Whiteman CD, Dorninger M, Pospichal B, Eisenbach S, Holzer AM, Weihs 663 P, Mursch-Radlgruber E, Baumann K. 2007, A sinkhole field experiment in the Eastern Alps, Bull. 664 Am. Meteorol. Soc., 88,701–716. 665 666 Tang ZY, Fang JY. 2006. Temperature variation along the northern and southern slopes of Mt. 667 Taibai. China. Agric. Forest. Meteorol. 139: 200-207. 668 669 Virtanen T, Neuvonen S, Nikula A. 1998. Modelling topoclimatic patterns of egg mortality of 670 Epirrita autumnata (Lepidoptera: Geometridae) with a geographical information system: 671 predictions for current climate and warmer climate scenarios. J. Appl. Ecol. 35: 311–322, DOI: 672 10.1046/j.1365-2664.1998.00299.x. 673 674 Vosper SB. 2004. Inversion effects on mountain lee waves. Quarterly Journal of the Royal 675 Meteorological Society, 130(600), 1723-1748. 676 677 Warner TT, Peterson RA, Treadon RE. 1997. A tutorial on lateral boundary conditions as a basic 678 and potentially serious limitation to regional numerical weather prediction. Bull. Amer. Met. 679 Soc.,78(11), 2599. 680 681 Whiteman CD. 1982. Breakup of Temperature Inversions in Deep Mountain Valleys: Part I. 682 Observations. J. Appl. Meteor. 21, 270–289. 683 684 Whiteman CD, Allwine KJ, Fritschen LJ, Orgill MM, Simpson JR. 1989. Deep Valley Radiation 685 and Surface Energy Budget Microclimates. Part I: Radiation. J. Appl. Meteor., 28, 414–426. 686 687 Whiteman CD, Bian X, Zhong S. 1999. Wintertime evolution of the temperature inversion in the 688 Colorado Plateau Basin, J. Appl. Meteorol., 38,1103–1117. 689 690 Whiteman CD, Eisenbach S, Pospichal B, Steinacker R. 2004, Comparison of Vertical Soundings 691 and Sidewall Air Temperature Measurements in a Small Alpine Basin. J. Appl. Meteor., 43, 1635–692 1647. 693 694 Whiteman CD, Hubbe JM, Shaw WJ. 2000. Evaluation of an inexpensive temperature datalogger 695 for meteorological applications. J. Atmos. Oceanic Technol. 17, 77- 81. 696 697 Whiteman CD, Hoch SW. 2014. Bingham mine cold-air pool structure and evolution, Final Report: 698 Dept of Atmospheric Sciences, University of Salt Lake City, Utah, 51 pp. 699 700 Xercavins A. 1985. Els climes del Pirineu Oriental: desde les terres Gironines fins a la Catalunya 701 Nord i Andorra, Documents d’Anàlisi Geogràfica, 7, 81-102 (In Catalan). 702
Page 18 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
703 Xue M, Droegemeier KK, Wong V. 2000. The Advanced Regional Prediction System (ARPS)–A 704 multi-scale nonhydrostatic atmospheric simulation and prediction model. Part I: Model dynamics 705 and verification. Met. Atmos. Phys. 75(3-4), 161-193. 706 707 Zängl G. 2005. Dynamical aspects of wintertime cold-air pools in an alpine valley system. Mon. 708 Wea. Rev.133: 2721–2740. 709 710 Zhong S, Whiteman CD, Bian X, Shaw WJ, Hubbe JM. 2001. Meteorological processes affecting 711 the evolution of a wintertime cold air pool in the Columbia Basin, Mon. Weather 712 Rev., 129(10),2600–2613. 713 714 7680 words approx 715 716 List of Table Captions 717 Table 1: Lapse rate statistics for Conflent (6 stations) and the whole dataset (39 stations). Units 718 are ºC/km. The frequencies of absolute inversions, rates shallower than 2, 5ºC/km and steeper 719 than 10ºC/km are listed in the last four columns in each half of the table. 720 721 Table 2: Classification of inversion type based on presence or absence of inversion layers in 722 each elevation band. Inversions are represented by positive gradients of the regression line fitted 723 to the temperatures in each band. The first column lists frequencies based on all observations, 724 the last column only using regression gradients significant at p
-
Figure 8. ARPS model simulations for the inversion event of 12 Dec 2012: 0000 UTC a) Plan 758 view of surface winds (arrows) and air temperatures (ºC) by colour shading. White lines 759 represent transects shown in other panels. Ellipses represent Cerdanya (left) and Conflent 760 (right), b) longitudinal transect along Cerdanya from SW to NE (dotted line): Winds and air 761 temperature. Bottom row panels c) to e) Malniu/Masella transect from NW to SE (connected 762 circles): c) Winds and air temperature, 0000 UTC d) Winds and air temperature, 0600 UTC e) 763 Winds and potential temperature (K), 0600 UTC. 764 765 Figure 9. Modelled vs observed temperatures at four locations during the inversion episode of 766 28-30 January 2013. a) Fon0, b) Les4, c) Val4, d) Val3. 767 768 Figure 10: ARPS model simulations for the inversion event of 30 Jan 2013: Panels same as in 769 Figure 8. 770
Page 20 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
Table 1: Lapse rate statistics for Conflent (6 stations) and the whole dataset (39 stations). Units are ºCkm
-1. The frequencies of absolute inversions, rates shallower than 2, 5ºCkm
-1 and steeper than
10ºCkm-1 are listed in the last four columns in each half of the table.
Conflent: n=6 Whole Cerdanya dataset: n=39 Month Mean
ºCkm-1
r2 >0 >-2 >-5 0 >-2 >-5
-
Table 2: Classification of inversion type based on presence or absence of inversion layers in each
elevation band. Inversions are represented by positive gradients of the regression line fitted to the
temperatures in each band. The first column lists frequencies based on all observations, the last
column only using regression gradients significant at p
-
Table 3: Mean bias statistics and RMSE for a selection of stations for each inversion modelling episode. Units are ºC. The last column represents the correlation between instantaneous model and observed temperatures over the whole period of the simulation.
Station Elevation Mean bias
Max bias
Min bias SD bias RMSE Correlation
Dec 2012
ºC ºC ºC ºC ºC
Cad1 2229 -3.48 3.80 -10.86 3.58 4.96 -0.133
Fon0 2381 -2.23 3.57 -9.92 3.66 4.25 0.608
Les4 1588 -1.13 5.41 -4.85 2.10 2.36 0.953
Mas4 1410 -0.12 4.13 -5.76 2.59 2.56 0.702
Val3 1015 6.53 12.72 -4.88 4.81 8.07 0.919
Val4 1109 3.77 13.29 -6.25 6.04 7.06 0.925
Jan 2013
Cad1 2229 -7.46 8.69 -13.46 4.68 8.78 0.678
Fon0 2381 -7.95 10.21 -16.49 5.71 9.75 0.683
Les4 1588 -5.63 2.07 -13.89 3.94 6.85 0.883
Mas4 1410 -0.45 4.36 -7.51 3.17 3.17 0.549
Val3 1015 6.05 15.83 -3.23 5.08 7.86 0.772
Val4 1109 3.76 15.22 -8.76 6.80 7.71 0.787
Page 23 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
Page 24 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
2 4 6 8 10 12Month
a) All stations
0
2
4
6
8
10
12
14
16
18
20
22
24Ti
me:
UTC
-12-11-10-9-8-7-6-5-4-3-2-101234
2 4 6 8 10 12Month
b) Low band (1900m)
0
2
4
6
8
10
12
14
16
18
20
22
24
Tim
e: U
TC
-12
-11
-10
-9
-8
-7
-6
-5-4
-3
-2
-1
0
1
2
3
4
Page 25 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
Page 26 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
Page 27 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
-15
-10
-50
510
11 Dec 0000 12 Dec 0000 13 Dec 0000Time
All Stations Low: 1900 m
Laps
e R
ate:
deg
C/k
ma: Inversion 11-13 Dec 2012
-15
-10
-50
510
28 Jan 0000 29 Jan 0000 30 Jan 0000 31 Jan 0000Time
All Stations Low: 1900 m
Laps
e R
ate:
deg
C/k
m
b: Strong Inversion 28-31 Jan 2013Page 28 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
Page 29 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
Page 30 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
Page 31 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
Page 32 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications
-
For Peer Review
Page 33 of 33
http://mc.manuscriptcentral.com/metapps
Meteorological Applications